Midjourney Demands Disney and Rivals Reveal Their Internal AI Use

Midjourney has filed a court motion demanding that Disney, Universal, and Warner Bros. disclose full details of how they use artificial intelligence internally. The three entertainment conglomerates sued the image-generation startup in 2025, claiming its systems unlawfully reproduce copyrighted characters. Midjourney maintains that training models on existing creative work falls under fair use.

The current clash turns on the scope of evidentiary discovery. A court previously ruled that the studios only need to produce records about generative AI when those tools yield final, consumer-facing video or still images. Midjourney is now challenging that restriction, arguing it lets the plaintiffs handpick files that support claims of financial damage while concealing their own internal development pipelines.

The startup's defense relies on establishing an industry norm: if studio teams train internal models on unlicensed source material to generate concept art or storyboards, training on copyrighted data cannot easily be framed as exceptional misconduct. As part of the same filing, Midjourney asked the court to force the studios to disclose every prompt their staff ran through the generator, rather than solely the queries that generated infringing outputs submitted as exhibits.

The studios' lead attorney, David Singer, described the request as a fishing expedition designed to distract from the core infringement claims. According to Singer, the studios are not attempting to ban AI technology or shutter Midjourney, but to halt unauthorized reproduction and derivative works. The court has yet to rule on whether to expand the disclosure orders.

Source: habr

Court filing: U.S. District Court Document

Secondary report: Variety